DESCRIPTION (adapted from the applicant's description): Many different RNA binding and catalytic activities can be isolated from pools of random sequences, and recent work suggests that the same will be true of proteins. It is proposed to use in vitro selection and directed evolution to explore ways for generating new molecular functions de novo from random sequences and by modification of existing functions. The relationship between information content and functional activity will be examined to gauge the difficulty of obtaining progressively more active molecules. The re-use of previously selected functions to evolve new functions will also be studied. The idea that proteins are better than RNA at evolving new functions from existing functions will also be tested.